Yeah - thanks Tesla23. I take your points. What is interesting is that one can come down to quite low sampling frequencies and the relationship for the RMS value still holds pretty well.

To the Ineffable All,

I believe Claude Shannon's sampling theorem says that the sampling frequency should be at least 2 times the highest sinusoidal frequency you wish to recover.

Ratch

The relationship holds for any frequencies, as long as they are not in a fixed phase relationship. If I chop a waveform with duty cycle D then the power drops by D and the rms amplitude by √D. If the chopping frequency is low then as long as you integrate over a long enough period you will find the relationship still holds. The spectrum may be all messed up, but the RMS level will still be reduced by √D.

 

Thanks everybody. Great help.